T1 · SYNTHETIC TRANSCRIPT
Name the first version
“MindBridge is a memory tool for AI agents. It should capture decisions and preferences so I do not have to repeat myself in every new chat.”
Fictional input · no private store connected
Interactive synthetic demo · no personal data
THE COMPANION LOOP
Remember change—not just facts.
Four fictional conversations show how MindBridge separates facts, inferences and confirmations—then updates its memory when the person changes direction.
Explore the memory loopWorkspaceProduct direction shift
GUIDED DEMO4conversations
1confirmed shift
1superseded identity
1safety rejection
conversationcandidateconfirmed memory
01 · MEMORY ADMISSION
Four days. One honest change of mind.
Pick a day to inspect what the system kept—and what it refused.
T3 · MEMORY UPDATE
2 durable memories written
- #201Product identity: memory tool for AI agents
- #202Core job: carry decisions and preferences across chats
02 · MEMORY RECEIPT
Don’t trust the answer. Inspect why it remembers.
Choose a question and inspect the current memory plus its historical evidence.
Answer grounded in 2 memories
temporal_queryMindBridge is now a reflective AI companion. Its Memory Core is the technical foundation that keeps long-term context traceable, editable, and aware of change.
[205]
Product: reflective AI companionAug 14 · cosine × time decay
0.972 · current[206]
Memory Core is infrastructureAug 14 · cosine × time decay
0.931 · currentWhat this page proves
Inspect the implementation This deterministic synthetic scenario is safe to explore and is not connected to Cassie’s private transcripts or memory store. The visible answers are fixtures; the underlying project implements the same T1/T2/T3, time decay, superseded history, REST and MCP boundaries.